How to Hire Data Scientists
We’ve worked directly with data scientists for the past seven years, and the role is evolving as quickly as enterprise data. In order to ride the wave of change, we ask 5 questions of every data science candidate at Coseer in order to gauge their skillset and potential fit.
What’s so valuable about data science? It’s a wide field, but above all else, it’s about uncovering insight. It’s harder than it seems. True insight isn’t obvious or easily-accessible. It’s surprising, counter-intuitive, and often hiding in unwelcoming places. It takes the left brain and the right brain, working together, to uncover.
This is what we ask data scientists to do for us.
We’ve found that open-ended interview questions work best to gauge candidates for a unique role like data scientist. Because we’re looking for a rare blend of tech savy, analytical thinking, and creativity, facilitating a mini-brainstorming sessions works best. We want to get the creative juices flowing and turn an interview into a real two-way dialogue. If we can get a glimpse into how a candidate thinks through a problem, we quickly get a sense of how he/she can uniquely contribute to Coseer and ultimately fit in with the team.
We go through the below exercise during a candidate’s evaluation:
- We start with a real world case study and ask about how the candidate will solve various related problems.
AI or Deep Learning is not the answer to every problem. In fact, most of the situations can be solved with much simpler approaches.
- We then pick a problem and ask the candidate to estimate how much data is necessary to train the model to solve the problem. This lets us assess whether the candidate knows basic algorithms, and also whether he/she has implemented an AI model before. This discussion inevitably goes into modelling, data structure, etc… It’s a great starting point for any direction you’d like to go.
- We ask what the candidate would do if only 50-90% of the required data is available.This helps us gauge basic principles, creativity, and understanding of real world situations. In the real world, few enterprise teams have all the data they need. In most cases, the solution will not be as accurate as expected in the case study.
- We then ask the candidate how they’d improve accuracy to reach acceptable standards. It is important to understand that oftentimes, a model is not the be-all-end-all. In most practical applications, models must be flexible. If the proposed solution is AI, the model must work with humans and other validation processes/sources to give acceptable results.
- Finally, we ask the candidate about the ROI of this process in the case study. This is the domain expertise “secret sauce”. We don’t expect a data science candidate to know all about the business side of things, but there are no isolated projects. Everything must live in real world at the end of the day. The best candidates have a solid understanding of the business concerns beneath their assigned projects, and they can keep metrics like ROI in mind as they go.
If you’re ready to embark on your journey into enterprise AI and want some deeper perspective into applications of data science, setup a meeting with our team.